Feature Selection Method Based on Partial Least Squares and Analysis of Traditional Chinese Medicine Data
Algorithm 1
LAPLS
Input: Dataset D
Output: LAPLS regression equation
Begin:
(1) Standardize the dataset D to get ;
i = 1
(2) While (the number of latent variables i has yet to reach satisfactory accuracy)
(a) Calculate the maximum eigenvalue of the matrix and its corresponding eigenvector ;
(b) Calculate the latent variables score vectors and based on the feature vector ;
(c) Calculate the load vector and the square of :,, and the residual information matrix and ;
End
(3) Solve the multiple regression equation and denormalize the regression coefficient: ;
(4) Construct the objective function in conjunction with the L1 regular term constraint: ;
(5) Use the coordinate descent method to iterate multiple times, solve the compressed regression coefficient , and construct a new regression equation ;